AI-200 Exam Guide & Practice Questions

A team runs a Linux custom container on Azure App Service, and the image is already stored in Azure Container Registry. The application must receive an API endpoint named API_BASE and a mode value named FEATURE_MODE. The team wants to change these values between deployment environments without editing or rebuilding the image.

Which configuration should the team use to pass the required values to the container?

• AAdd API_BASE and FEATURE_MODE as App Service app settings
• BAdd API_BASE and FEATURE_MODE to the Dockerfile after the image has been built
• CAdd API_BASE and FEATURE_MODE to the deployment slot name
• DAdd API_BASE and FEATURE_MODE as labels on the Azure Container Registry image

A development team has completed a new version of a web service and has a working Dockerfile. Its AKS cluster already has access to the team's Azure Container Registry, and the deployment manifest is maintained in source control. The new version must use the tag 2.4.0, and the team wants the cluster to run that image rather than the currently deployed version.

Order the steps to deploy the application version to AKS.

Tap the steps in the order you would perform them.

• Apply the manifest to the AKS cluster
• Build the container image and assign the intended version tag
• Update the manifest to reference the tagged image in Azure Container Registry
• Push the tagged image to Azure Container Registry

A background worker runs as an Azure Container App and processes messages from an Azure Service Bus queue. The current configuration keeps one replica running even when the queue is empty. The team wants the worker to scale to zero when idle, begin scaling when the queue has about 20 messages per replica, and never exceed 12 replicas during a burst.

Select the appropriate value in each dropdown to meet the worker scaling requirements.

Choose the correct option for each statement.

A recommendation service stores user activity in an Azure Cosmos DB for NoSQL container. A worker must process newly created and updated items, and several worker instances may run at the same time. The team has created a separate container for processor state and wants the application code to handle each batch of changes.

Match each component to its role in the change feed processor.

Match each item to the correct target.

An inference service stores request history in Azure Database for PostgreSQL Flexible Server. A frequent endpoint runs a query that filters rows by customer_id, orders the matching rows by created_at descending, and returns the newest 20 records. The table is growing quickly, and the database team wants one index that supports both the filter and sort.

Which index should the team create to best support this query pattern?

• AA separate B-tree index on customer_id and another B-tree index on created_at
• BA composite B-tree index on customer_id and created_at in descending order
• CA GIN index on customer_id and created_at
• DA B-tree index on created_at in descending order followed by customer_id

A support assistant stores generated embeddings in an Azure Cosmos DB for NoSQL container alongside article identifiers and text. The application must retrieve semantically similar articles from the same container. The team is defining the container configuration and query, and needs to distinguish vector-specific settings from ordinary indexing and text search features.

Select the appropriate configuration or query function for each statement.

Choose the correct option for each statement.

A document processing platform publishes custom events to an Azure Event Grid topic whenever a review status changes. The approval workflow should receive events only when the event data field status is approved. Other workflows use the same topic and must continue receiving events according to their own rules. The team wants to avoid invoking the approval function for irrelevant events.

Which approach should the team use to deliver each qualifying event only to the required workflow?

• ACreate an Event Grid subscription with an advanced filter on the event data field data.status for the value approved.
• BCreate a Service Bus topic subscription with a SQL filter on data.status.
• CDeliver every event to the function and discard events with other status values in the function code.
• DConfigure a storage queue visibility timeout to accept only events with an approved status.

An AI inference platform uses Azure Service Bus for work coordination. Review notifications must be delivered independently to the audit service and the reporting service. Inference requests should be shared across a pool of workers, with only one worker processing each request. The operations team also needs failed requests isolated after repeated delivery attempts, while successfully processed requests must not be delivered again.

Match each requirement to the Service Bus feature or action that fits it.

Match each item to the correct target.

A partner application submits invoice data to an Azure Functions endpoint using HTTP POST. The function validates each request and then places a work item on an existing Service Bus queue for asynchronous processing. The team wants the function to use bindings rather than create a Service Bus client in application code. The function app already has a connection setting named BillingBus.

Select the appropriate function trigger and binding configuration for each statement.

Choose the correct option for each statement.

A Function App calls a partner service with a credential stored in Azure Key Vault. Its system-assigned managed identity already has the Key Vault Secrets User role. A scheduled rotation process creates a new version of the secret, and the application must use the current value without placing credentials in deployment settings or requiring a redeployment.

Which change best meets the requirements for retrieving the rotated credential?

• AUse the managed identity with the Key Vault client library to retrieve the secret by name, then refresh the cached value after rotation.
• BStore each new secret value as a plain value in Azure App Configuration and update it during rotation.
• CPlace the secret value in the Function App settings and restart the app whenever the credential rotates.
• DUse the secret URI as the credential value so the application does not need to access Key Vault.

A support team uses Application Insights to investigate a rise in API failures. The team needs a ranked list of operation names for failed requests from the last 30 minutes. The query should count requests rather than return individual records, and the analyst wants the operation with the highest failure count at the top.

Order the steps to produce a KQL result showing the operation names with the most failed requests during the last 30 minutes.

Tap the steps in the order you would perform them.

• Order the results by FailureCount in descending order.
• Filter for requests where success equals false.
• Group by operation_Name and count the failures as FailureCount.
• Start with the requests table.
• Filter to timestamps later than ago(30m).

A team is instrumenting two .NET services that call each other over HTTP. The traces currently appear as unrelated operations, and engineers cannot reliably identify which service produced each span. The application must send telemetry to Azure Monitor, and the team wants to fix the instrumentation rather than infer service relationships later from log queries.

Choose the appropriate OpenTelemetry configuration for each tracing requirement.

Choose the correct option for each statement.